On Several Algorithms For Blind Channel Identification And Equalization

Blind identification and equalization is the pivotal technology in communicationsignal precessing, the main work of this dissertation is on this topic. At the first, thetheoretic basis is expatiated. Then we detailedly discuss how to design a criteria forblind deconvolution of nonminimum phase systems. And then we study the subspacemethods and the time domain approach for blind identification and equalization, whichare based on second-order statistics. In the research, lots of work of simulation isconcerned. At the last, a new recursive algorithm for blind identification andequalization is proposed. The algorithm exploits the cyclostationarity of oversampledcommunication signals to achieve identification and equalization of possiblynonminimum phase channels. Compared with other blind equalization algorithms whichonly exploit one or two eigenmatrix, this algorithm utilizes a set of eigenmatrices, sothat the performance is improved. By solving the cost function to obtain the bestsolution, the recursive algorithm is given which can obtain all the eigenvectors one byone. Further, the analytic solution is provided. Simulations show that algorithm here iseffective even under the condition of low signal-to-noise ratio (SNR) which canovercome the shortcomings of conventional algorithms. In addition, our algorithm haslow complexity of computation.